Inside GM's State-of-the-Art Powertrain Engineering Center

Photos courtesy of General Motors

The engines in today’s GM pickup trucks have never faced stiffer competition, but there’s not much that can be done about their design or performance except to look at the future, which GM is actively shaping inside its state-of-the-art Powertrain Engineering Development Center in Pontiac, Mich.

The $463 million benchmark facility opened in early 2008 next to the company’s Global Powertrain headquarters. It’s the crown jewel of more than 30 years of planning and restructuring.

Until the 1980s, GM’s brand divisions were managed as a confederacy. Chevrolet, Cadillac, Buick, Oldsmobile and Pontiac all maintained separate headquarters and engineering staffs that only came together on the bottom line of the company’s financial statements. It wasn’t unusual for similar vehicles, like the Chevrolet Camaro and Pontiac Firebird, to use V-8 or V-6 powertrains with designs and parts that didn’t share much in common except for their intended purpose and maybe a few bolts.

Industry rivalry changed all of that. As Japanese cars became popular and Ford and Chrysler grabbed market share from GM with new products, GM consolidated divisional engineering teams into shared resources that would design fewer engines at less cost with increased reliability.

In 1984, five engineering development teams became two with the creation of the Buick-Oldsmobile-Cadillac and Chevrolet-Pontiac-GM of Canada Powertrain divisions. In 1990, they merged into a single entity, General Motors Engine Division. In 1991, GMED and Hydra-Matic transmission groups were reorganized to create GM Powertrain. Later that year, the Central Foundry Division and Advance Engineering were merged into the division. In 1997, the formation of a GM global powertrain organization was announced, encompassing all of GM's powertrain engineering and manufacturing activities outside of North America.

Dyno cells are built to test everything from a standalone engine to a full powertrain. Here, a 5.3-liter V-8 engine is tested with its transmission.

Over time, multiple GM powertrain development centers around Detroit were phased out. GM opened the doors to the Powertrain Engineering Development Center in 2008, on the site of the old Pontiac Motor Division.

But it wasn’t just GM’s disjointed powertrain organization that changed. Three decades ago, computers were quickly finding their way into automotive development. They became invaluable because engineers could save time (and money) by designing and modifying parts on a screen instead of by hand.

Computer aided design was soon supplemented with computer simulation, which tested parts performance virtually before a physical copy was ever fabricated and tested in the real world. Modeling and simulation became more sophisticated as the lessons learned from the previous generation of car or truck were applied to the next generation.

GM slowly transformed its powertrain development process from one that designed and improved engines and transmissions based on real-world trial and error over thousands of vehicles and miles into a modern methodology that has more in common with software development than old-school powertrain hacking.

“If you can move work from the road to the lab, that’s fast. If you can move from the lab to math (data), that’s even faster. That’s our philosophy,” said Radu Theyyunni, GM powertrain engineering group manager for system design and analysis. In GM engineering speak, it’s called RLM: road to lab to math.

Theyyunni’s team is brought into a vehicle’s development very early in the lifecycle to help determine its performance and fuel economy targets. If a pickup truck requires a new powertrain, the mill’s displacement, architecture (diesel, gas, direct injection, turbocharged), transmission and axle ratio are estimated, and software is used to calculate the engine’s bore, stroke and crank size to model its power and torque characteristics. All of this happens before a single engine drawing or design has been created. This is where the rubber meets the road for Theyyunni and his number crunchers.

“Our goal is to use simulation methods to make sure that the first time we [physically] build a powertrain, it’s the optimal design,” Theyyunni said. “If you look back to before 2000, we’d design stuff and test it. It was a design and test philosophy. If a part broke, we’d go back and redesign and test it. There would be a postmortem where we’d be asked why [it] broke, and we [could provide a reason] but we weren’t driving the design. It was more like a support role, and each build could take six to eight months. What we’ve done in the last 10 years [is] move to math-based development.”

Today, powertrain analysis and design is done in short cycles, using powerful computers to develop a new engine.

Friction, temperature, emissions, fuel economy and other key traits are modeled and simulated multiple times to optimize component performance long before the first physical part of an engine is tested on a dyno.

“It’s like a zipper,” Theyyunni said. “Now, we do hundreds of [analysis and design] iterations [using computers] depending on what the component is. Then we test in a virtual validation using characteristics modeled in math, such as the ports [which determine how well an engine can breathe], block size and seal and gasket sizing. Each part in the [engine’s] bill of design and materials uses [math-based] rule sets that we’ve developed over time that have become our best practices. That’s what we use to create our simulations. Once you build your hardware, it’s too late to make a [design] change.

Engines are quickly moved in and out of dyno cells on pallets that float on a cushion of air like an air hockey puck.

“Back in the old days, we could always tell you why something broke,” Theyyunni said. “But today when I tell you up front it’s going to break here, it’s a whole new level of responsibility. [Our role] is more challenging, but it’s also more fun.”

The lessons learned from this iterative approach have helped GM improve each new generation of engines and have shortened the time required to develop them. Novel data and lessons learned from real-world testing that don’t agree with existing simulations are turned into new math-based rules that are factored into the next round of virtual validation testing.

“That’s where continuous improvement comes from, and it’s how we get better over time,” Theyyunni said. “Analysis, design and testing are linked together. Some of the designs are baked into our software so if a designer tries to build something out of whack, he’ll get a warning message that he’s violating his part guidelines.”

Depending on complexity and scope, some engine activity modeling can require up to three days of non-stop computing time using more than 100 CPUs to render three minutes of enhanced video footage that can be reviewed and understood by a powertrain engineer. That’s still shorter than what used to require weeks or months testing an actual engine on a dyno, and failing an engine in a computer is cheaper than failing an engine on a dyno.

Part of Theyyunni’s recent modeling efforts were the analysis and design of GM’s next-generation small-block V-8 that will help power all-new Chevy and GMC pickups by 2013 (which we expect will be 2014 model year trucks). The engines will feature aluminum engine blocks, direct injection and an advanced combustion system.

“We looked at hundreds of combinations of ports, chambers, cam designs and injector designs,” Theyyunni said.

In a non-fueled test cell, a transmission is coupled to an electric motor that stands-in for a gasoline-powered engine.

The latest version of the 6.6-liter Duramax V-8 diesel for the 2011 Chevy and GMC heavy-duty pickups required more than 500 piston bowl designs – a special shape carved in the top of the piston that helps control the way air swirls above the piston during combustion – before the optimal shape was created and rendered in metal.

Finally, after all the analysis, design and simulation are complete, a cross-functional group of engineers and managers reviews the results. If approved, the virtual engine becomes reality within walking distance of the computers it was designed on, and then it’s on to dyno testing or shipped to GM’s Milford Proving Grounds to be installed in a vehicle.

The 450,000-square-foot Powertrain Engineering and Development Center has two test wings that house 85 dynamometer test cells. It also has 100 powertrain component test stands and a powertrain engineering factory where prototype engines — from gasoline to electric — are built and assembled.

Steve Nash, the center’s engineering development manager, walked us around.

In engine dyno cells, test engines are subjected to different loads and environmental conditions using various calibrations designed to help the engine adapt to workload and climate while ensuring the engine doesn’t break. Valvetrain performance is also measured and checked against design specs. What used to require intrusive sensors installed on valves – which added mass and changed a valve’s shape – to measure engine timing is now done passively with lasers that return significantly more accurate data.

To help reduce the costs and risks associated with hot and cold weather simulations, rather than heat or cool entire dyno cells, environment boxes are lowered into place around engines or entire powertrains to simulate temperature extremes, from minus 40 degrees to more than 130 degrees.

Thermal oxidizers destroy at least 96 percent of carbon monoxide produced by test engines before the exhaust is released.

Non-firing dyno cells are used to spin up and test transmissions without the need for a conventional engine. Small low-inertia AC electric motors that can be programmed to simulate the firing pulses of almost any kind of internal combustion engine, from four-cylinder to V-10, are mated directly to the experimental gearbox’s torque converter.

“In the old days, we used to always have to have an engine and transmission to do dyno work,” Nash said. “The prime mover was the engine, and the transmission factory would always say we have to wait for the engine before we could test the transmission. In this facility, we’ve decoupled that. We don’t need an engine to start transmission calibrations and development. It’s all about being fast to market.”

Dynoed transmissions can also be paired with test or production rear differentials to simulate drivetrain load. To speed up the powertrain test process even more, the entire rig — with or without an engine — can be tilted dynamically in a way that simulates driving on an actual road or highway to start dialing in transmission calibrations before a future engine is ever fabricated and paired with the gearbox.

“Any road course we have at Milford can be simulated on our tilt stands, including compound angles [which simulate stopping on an angle] and up to 1.3 g [through a turn],” Nash said. “If there’s an issue with a customer’s engine on a specific road course, we can map [the road], bring [the model] in here and rerun it.”

In 63 percent calibration dyno cells, the entire powertrain and drivetrain is brought together physically for the first time. Calibration data from the test is provided to vehicle engineers at Milford, where they can use that information to get a jump-start on making further adjustments and changes in test vehicles running in the real world.

Another innovative technique used to reduce testing time is the plug-and-play movement of engines into dyno cells on air-floated pallets. All of the cables and instrumentation needed to record data during dyno testing that used to be part of the test chamber is rigged to the pallet and engine before the entire unit enters a cell. Changeover procedures that used to require a full day connecting an engine to sensors and fluid feed lines can be now be completed in as little as 20 minutes. This also allows the dyno chamber to have maximum uptime for improved engine testing productivity, and fewer dyno chambers are needed to manage the work. A single person can move the air-cushioned pallets that tip the scales at 3,000 pounds like a giant air hockey puck.

Powertrain engineers and dyno technicians sit side-by-side outside each test cell during testing.

“We always want to make sure our test cells are running,” Nash said. “And we never have to change the test cell if an engineer comes along and says they need to collect more data. Before, they would screw up our [dyno cell] template. Now, the test cell always stays the same.”

Innovative ideas have also been implemented to manage emergencies. Instead of using carbon dioxide to put out a fire, high-pressure water mist is used to smother flames so rescuers or injured people don’t risk suffocation. Dyno chamber roofs are also engineered at an angle with blow-off panels in the event of an explosion.

All of the engine exhaust created during dyno testing is captured on site and funneled to four large thermal oxidizers that scrub and clean the exhaust to remove up to 98 percent of harmful carbon monoxide and soot before the exhaust is released into the air.

Up to 250,000 gallons of fuel is stored on site, and it takes one large refueling tank truck a day, like you’d see at a gas station, to keep things running. There are 27 fuel types, including all gasoline octanes, diesel, biodiesel, ethanol, methanol, compressed natural gas and liquefied petroleum gas.

All of this adds up to one of the most advanced powertrain facilities on the planet that’s designing and creating GM’s future, one engine at a time.